Early Methods for Detecting Adversarial Images

2016/08/01 by Dan Hendrycks, Hendrycks, Dan, Kevin Gimpel +1 · 2 citations
Biochemistry, Genetics and Molecular Biology · Computer Science · #Adversarial Robustness in Machine Learning #Anomaly Detection Techniques and Applications #Cell Image Analysis Techniques #Computer Vision and Pattern Recognition (cs.CV) #Cryptography and Security (cs.CR) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Neural and Evolutionary Computing (cs.NE)

paper · pdf · doi:10.48550/arxiv.1608.00530

openalex publication_date 2016/08/01 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Many machine learning classifiers are vulnerable to adversarial perturbations. An adversarial perturbation modifies an input to change a classifier's prediction without causing the input to seem substantially different to human perception. We deploy three methods to detect adversarial images. Adversaries trying to bypass our detectors must make the adversarial image less pathological or they will fail trying. Our best detection method reveals that adversarial images place abnormal emphasis on the lower-ranked principal components from PCA. Other detectors and a colorful saliency map are in an appendix.

Citations

• Sparse coding with an overcomplete basis set: A strategy employed by V1?
• Transferability in Machine Learning: from Phenomena to Black-Box Attacks using Adversarial Samples

Cited by

• Sparse Representations Improve Adversarial Robustness of Neural Network Classifiers
• Detecting Adversaries, yet Faltering to Noise? Leveraging Conditional\n Variational AutoEncoders for Adversary Detection in the Presence of Noisy\n Images

Related

• Detecting Adversarial Perturbations with Saliency
• On Detecting Adversarial Perturbations
• Adversarial Detector with Robust Classifier
• A general metric for identifying adversarial images
• Adversarial Examples Are a Natural Consequence of Test Error in Noise